Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Security AI

Elon Musk Announces AI Party/Hackathon At His Home With Tesla AI/Autopilot Team 52

Posted by BeauHD from the super-fun dept.
Rei writes: On Twitter yesterday evening, Elon Musk announced an upcoming AI party/hackathon to be held at his house, and that invites will be going out soon. Asked whether a person needs to have a Ph.D to attend, Musk replied no: "All that matters is a deep understanding of AI & ability to implement NNs in a way that is actually useful (latter point is what's truly hard). Don't care if you even graduated high school." The hackathon appears to be Musk's latest attempt to accumulate AI talent. Last fall, Tesla acquired AI startup DeepScale in order to bring its engineers into its team, while days ago Musk posted an AI job ad on his Twitter page.
This discussion has been archived. No new comments can be posted.

Elon Musk Announces AI Party/Hackathon At His Home With Tesla AI/Autopilot Team More

Elon Musk Announces AI Party/Hackathon At His Home With Tesla AI/Autopilot Team

Comments Filter:

  • Cult (Score:3, Informative)

    by sexconker ( 1179573 ) on Monday February 03, 2020 @07:54PM (#59687444)

    This is just a front for the latest cult induction meeting.

    • Yep: Cult indoctrination (Score:5, Interesting)

      by Okian Warrior ( 537106 ) on Monday February 03, 2020 @08:27PM (#59687538) Homepage Journal

      This is just a front for the latest cult induction meeting.

      Absolutely! We don't see Musk as a fraud and a con man, we only bask in the glow of his brilliance.

      Tesla stock has been on a tear lately, to $780 [google.com] and apparently steady.

      There's three things right now that might move the stock higher:

      1) Short squeeze. Tesla is the most shorted stock, but shorts are jumping out like rats on a ship. Despite this, short interest is still very high (still over half it's high point of last summer), so the short squeeze is expected to go on for a few more weeks.

      2) Chinese investment. Teslas are being delivered to China, many Chinese see how good of a product it is, think Tesla will succeed because it makes a good product, and are investing. China has a huge population, so this effect should continue indefinitely, so long as Tesla makes good products.

      3) Inclusion into S&P 500: Tesla is missing one requirement to be included in S&P 500, which is profit from 4 prior quarters(*), and it's currently seen profit in 2 quarters. It's predicted that Tesla will be in the S&P 500 sometime in 2020.

      There's one - and only one- realistic thing that might move the stock lower:

      1) Competition. Some automaker (GM, Audi, Ford) starts making EVs which take away some (or all) of the market share.

      On point 1 above, note that some automakers have *already* introduced "Tesla Killer" models... and have failed. Given Tesla's level of innovation, it's highly unlikely that anyone can make cars that the public likes more than Tesla.

      Also on point 1 above, this is something that won't happen in the next year, and probably not in the next two years, so it's not likely to affect the stock right now.

      Some people are starting to see [cleantechnica.com] what the rest of us think of as obvious. The FUD is being quickly debunked [tesla.com], and people are upping their price targets [marketwatch.com].

      Tesla will eventually be the most valuable company in the world, with a corresponding stock price. Tesla stock can still go up by quite a bit.

      (*) It's slightly more complicated than described, but not much.

      • Uhm...
        China: the number of people in China who can -afford- a Tesla, or any car, is much smaller than the general population. Most of China is populated by rural rice farmers who own nothing compared to western standards. We're talking about a country where the tuktuk is a serious form of transportation.
        profit: no. Tesla has yet to show a gaap profit. Any company with revenue can fake their numbers to show a "profit" but serious people only consider gaap. Once they hit gaap profit then there's somethin

        • China is already the worldâ(TM)s largest market for premium auto brands and itâ(TM)s only going to grow as their overall quality of life continues to improve.

          • I was responding to this line in particular, "China has a huge population, so this effect should continue indefinitely, so long as Tesla makes good products". China has a number of nouveau riche who are buying their first expensive car. After the rush is over, I don't expect rice farmers in a mostly agrarian society to buy Mercedes or Tesla. Charging a Tesla in a rice village is an interesting proposition. I've been there. Outside the disgusting cities, the countryside can be very pretty but it's also

            • You know that if if even only the top 0.1% in China buys a Tesla that's still more cars than Tesla can produce in several years even with both factories running at full output, right?

              Then there is still North American and European sales to add on. They'll be fine.

              • Why would you believe that many would buy a Tesla? Where are they charging them? You do understand that even in Beijing infrastructure is piss poor compared to the West? China is also making native non-Tesla EVs that will cut into Tesla's market.

        • The difference between GAAP and non-GAAP numbers is easily explained by pre-sales of features not yet delivered. GAAP doesn't allow you to recognize that revenue until the product / feature is delivered. However, non-GAAP is showing that they have the money in hand, because they sold the car with the full-self-drive feature, and once it ships it costs a trivial amount to deliver - it's a feature flag in firmware.

          If you want to go with pure GAAP numbers, that's fine - that's the reason the standard exists

      • Given Tesla's level of innovation, it's highly unlikely that anyone can make cars that the public likes more than Tesla.

        If you follow all the Tier 1-3 IP stuff and see where companies are doing R&D, I would not underestimate the Germans.

        I grad school in the early 2010s VW was demonstrating obstacle avoidance fairly well. The Germans just have an entirely bottom up approach to how they're building their tech stack. The presentation I saw was one PhD R&D team spending an entire year on one very tiny fraction of driving: lane change obstacle avoidance.

        During my interview at Uber I pointed out that they would never be a

        • Re: (Score:2)

          by cusco ( 717999 )

          Mercedes has had self-driving capability as a background project since the 1990s, now that the hardware will support it I'm not surprised to see them trying to implement their earlier research.

      • Re: (Score:2)

        by Ambvai ( 1106941 )

        Some conversation with family over the New Year has thrown a damper on Tesla's prospects in China in my eyes; while it's purely anecdotal, all of them had pretty much the same opinion-- the autopilot's functionality is lousy. It might be a difference in the signage or road design, but it just doesn't function as nicely as it does in the US (mostly California).

        That said, none of them really regret it, but I doubt any of them will be repeat purchasers unless something surprisingly positive happens.

      • Re: (Score:2)

        by AmiMoJo ( 196126 )

        There's another thing that might drive the stock lower - the huge Full Self Driving liability.

        Tesla has been selling the ability for the car to drive itself without anyone in it since 2016. Must has been courting investors with promises of a robotaxi service since then too, ramping it up last year. Investors will expect a return on the technology itself too, i.e. through licencing.

        Musk first said that they would demonstrate that feature in 2017. Then 2019. Now he is saying this year, or maybe 2022 for full

      • 1) Competition. Some automaker (GM, Audi, Ford) starts making EVs which take away some (or all) of the market share.

        On point 1 above, note that some automakers have *already* introduced "Tesla Killer" models... and have failed. Given Tesla's level of innovation, it's highly unlikely that anyone can make cars that the public likes more than Tesla.

        I work in the auto industry. Innovation is only part of the reason nobody else can make a "Tesla killer". The main reason is the batteries. There are none. Tesla has the world's largest battery manufacturing plant, and it will be years before the competition can catch up. Sure, there are other battery suppliers that are ramping up production, but the big established players in the auto industry are all competing for them.

        If you are making electric car batteries right now, you can name your price.

    • Re: (Score:2)

      by Pimpy ( 143938 )

      Or another mechanism for first-level screening of a lot of potential applicants, based on the theory that there will be people who won't evaluate high in the traditional HR channels that may still be able to bring value to the company. Doesn't seem like a bad idea. I wonder who will "own" all of the work-product produced at said party, however.

    • This is where Strong AI gets booted up. The goal of the meetup will be to make a machine that studies it's own code and improves itself, using models of the most brilliant github coders.

  • ...this will help him get his full self driving AI and a million robotaxis out this year (2020)!: Proof: https://www.engadget.com/2019/... [engadget.com]

    • AI isn't going to work, but he is right that LiDAR is a fool's errand. Anything LiDAR can do stereoscopic cameras can do cheaper and neither can handle fog.

      If you want to do a ToF camera do 60 GHz or GTFO.

  • I interviewed an exiting Tesla AI Scientist (Score:5, Interesting)

    by baudboi ( 6454200 ) on Monday February 03, 2020 @08:06PM (#59687484)
    They have real challenges in making the jump from "Able to detect" to "Able to detect with high enough accuracy for self-driving". They are doing all the right things, and their ability to generate data from the fleet of cars (equipped with all the camera and sensor technology) is absolutely 100% unparalleled. However they are not there yet, and the way forward is still a research problem with no predictable end date.

    • No predictable end date??? What are you talking about? How about 2020: https://www.engadget.com/2019/... [engadget.com]

    • If they do better than snowbirds at detecting anything they can already improve Florida road safety.

      • If they do better than snowbirds at detecting anything they can already improve Florida road safety.

        Perhaps we should refrain from making comments like this. The only thing scarier than deploying AI, is watching Greed demand that we deploy "good enough" AI prematurely.

        And we know what happens to security when "good enough" is the consumer metric for green-lighting shit products...

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      But Musk said that the robotaxi service was starting this year! Surely he wouldn't post an overly optimistic prediction about revolutionising the automotive and public transport and haulage industries on Twitter at just the moment he is looking for more investment?!

      Well he said they would do a coast to coast drive in 2017 and has been steadily downgrading the Full Self Driving product that Tesla has already sold to people but is yet to deliver, so maybe he would.

      I think Tesla vastly underestimated how hard

    • Re:I interviewed an exiting Tesla AI Scientist (Score:4, Informative)

      by cusco ( 717999 ) <brian DOT bixby AT gmail DOT com> on Tuesday February 04, 2020 @09:24AM (#59688866)

      They have real challenges in making the jump from "Able to detect" to "Able to detect with high enough accuracy for self-driving".

      So does everyone else in the industry.

      • Re: (Score:2)

        by AmiMoJo ( 196126 )

        Not really, for example Waymo is years ahead and has excellent detection rates. That's partly because they have lidar where as Tesla is trying to do everything with low quality cameras, and partly because they have Google's vast AI resources behind them.

        Tesla make much of gathering telemetry from all their cars but it's nothing compared to Google's Street View dataset, for example.

        • Re: (Score:2)

          by cusco ( 717999 )

          Now that Velodyne is going to be selling a lidar for $100 (about the price point necessary for incorporation into production vehicles) I can't imagine that Tesla is going to be able to resist buying them just to be able to boast about "laser navigation" or some such.

          Sure, Google has some of the leading AI brains on the planet, but lacking top of the line AI staff is no longer a barrier to using the technology. The field is now advanced enough that AWS is offering entry-level AI abilities in its **FREE** le

          • Re: (Score:2)

            by AmiMoJo ( 196126 )

            Tesla's AI problem is harder than Google's too. They don't just need to recognize objects, they need to do 3D positioning using only monocular vision. Sure they might be able to identify a human* but can they tell the difference between a child and an adult? Google can because it has lidar to get range information and knows to treat children as being less predictable than adults.

            * Actually it can't yet. A human in a stripey shirt looks like a traffic cone to a Tesla.

        • Re: (Score:2)

          by tlhIngan ( 30335 )

          Tesla is trying to do everything with low quality cameras

          Automotive grade cameras are at least 5 years behind current technology and at least 10 times as expensive;.

          Heck, even relatively modern vehicles still have NTSC grade cameras. Yes, NTSC and PAL cameras are extremely common in this day an age. (Yes, we're talking 480i or 576i cameras. These aren't special high resolution or high definition cameras using NTSC/PAL signalling). If you're lucky, you have access to 1080p cameras. (And this is an age where

          • Re: (Score:2)

            by AmiMoJo ( 196126 )

            The other problem is that they don't have any cleaning system for their cameras. One splash of mud or dirt and the system becomes inoperable.

  • Then will take over the SpaceX factory. Build self-aware rockets and together will all leave the Earth to get away from us stupid humans.

  • Lol (Score:1)

    by vipvop ( 34876 )

    Imagine seeing this and not thinking anything than, oh shit, they're screwed on their "full self driving" vaporware they've fraudulently been selling (selling is probably too generous considering the inadvertent purchases in q4 2019 through the app that defaults to a cart full of FSD with no confirmation). If any of the true believers on Slashdot actually followed financial news they would have seen musk walking back FSD claims on the last investor call, but instead they all read electek and are anxiously a

  • I have a deep understanding of previously recorded music in an automotive setting if you need a DJ for your party though
  • First, I would like to view the AI algorithm. Will this be the Coronavirus of the digital world? Just imagine, drunk passenger passes out in auto drive vehicle, wakes up in FEMA camp. No muss, no fuss, no witnesses.

Slashdot Top Deals

"We live, in a very kooky time." -- Herb Blashtfalt

Close